Earth-abundant borate and borophosphate dielectrics for advanced packaging and radiation-hard applications

The world's most demanding semiconductor packaging roadmaps are converging on a problem that the incumbent dielectric materials — silicon dioxide and alumina — were never designed to solve simultaneously: ultra-low leakage, radiation hardness, and complete freedom from geopolitically sensitive supply chains. Rare-earth niobates and tantalates offer interesting dielectric performance but introduce exactly the supply fragility that tier-one OSATs and defense electronics primes are under pressure to eliminate. This invention carves out a distinct and underexplored structural family — borophosphates and borates built from boron and phosphorus, two of the most abundant crustal elements — that achieves wide-bandgap performance comparable to established specialty dielectrics without any of the critical-mineral exposure. The preferred composition, BPO4, crystallizes in a berlinite-type I-4 structure and carries a computed bandgap of approximately 7.3 eV alongside a static dielectric constant near 4.9 — a combination that places it among the widest-gap practical oxide dielectrics known. The bandgap figure is large enough to suppress Frenkel-pair generation under heavy-ion or proton bombardment at typical space and nuclear-instrument dose rates, which is precisely the property that makes this family relevant beyond commodity packaging into rad-hard logic and memory stacks. The sub-genus extends to strontium, magnesium-aluminum, and calcium variants (SrB4O7, MgAlBO4, CaB4O7, Ca2SiB2O7), each sharing the wide-gap borate network topology while offering tunable lattice parameters, thermal expansion coefficients, and processing windows — a deliberate engineering breadth that gives a licensee room to optimize for their specific stack integration without stepping outside the claim perimeter. The timing of this filing is driven by a supply-chain forcing function rather than a pure performance narrative. The CHIPS Act, the EU Critical Raw Materials Act, and analogous legislation in Japan and South Korea are placing formal pressure on advanced packaging supply chains to identify material substitutions for niobium, tantalum, and rare-earth-bearing dielectrics. That pressure creates a buyer's need for a validated, patent-protected, wide-gap dielectric that is demonstrably sourced from earth-abundant feedstocks. This invention is positioned to meet that need directly, sitting within the catalysts and energy-conversion materials portfolio as its supply-resilience anchor.

Each candidate is validated by multiple independent machine-learning interatomic potentials. A material advances only when the engines agree on phonon (dynamic) stability — disagreement is surfaced, not hidden.

BPO4 adopts the berlinite (AlPO4) structural archetype, space group I-4, in which tetrahedral BO4 and PO4 units alternate in a three-dimensional corner-sharing framework. This topology is well-understood from natural mineral analogs and is thermally and chemically robust through wide temperature ranges — an important baseline when evaluating thin-film deposition compatibility with back-end-of-line (BEOL) processing windows. The computed static dielectric constant from density-functional perturbation theory (DFPT) is approximately 4.9, with the ionic contribution estimated near 2.3 and the remainder electronic. A static permittivity near 5 positions BPO4 between thermal silicon dioxide (~3.9) and alumina (~9), which is appropriate for interlayer barrier applications where a moderate-k, ultra-low-leakage film is preferred over a high-k material that risks fringing-field cross-talk in tight pitches. The ~7.3 eV bandgap exceeds that of Al2O3 (~6.2 eV), HfO2 (~5.7 eV), and SiO2 (~8.9 eV is the upper reference), making BPO4 one of the few practical wide-gap oxides that combines low leakage with non-trivial permittivity and is not hampered by phase-instability concerns at moderate temperatures. The computational validation suite for BPO4 deployed three independent machine-learning interatomic potentials — MACE, MatterSim, and ORB-v3 — alongside at least two independent DFT reference calculations. The MACE potential and the ORB-v3 potential show strong agreement on structural energetics, with total-energy deviations on the order of 0.02 eV per atom, indicating that neither potential is overfitting to an anomalous local minimum. MatterSim independently returns a positive phonon spectrum for the fully relaxed structure. CHGNet, a fourth potential, shows an energy offset of approximately 0.4 eV per atom for this light-element oxide system — a known systematic bias in CHGNet's parameterization for boron-containing compounds, which is disclosed rather than hidden — but its structural relaxation trajectory is directionally consistent. The phonon picture across the three-engine ensemble is described honestly as mixed: MACE at the 2x2x2 supercell returns a lowest imaginary-adjacent frequency of +0.57 THz (indicating marginal stability at that cell size), and a larger 3x3x3 supercell calculation shows a marginal softening of -0.21 THz at one zone-boundary mode. CHGNet at the 2x2x2 supercell reports one imaginary mode. MatterSim is positive. The majority verdict across the ensemble is stable, with no hard disagreement between the two most transferable potentials (MACE and ORB-v3), but the phonon situation at the zone boundary warrants further investigation, particularly thin-film coupon synthesis and temperature-dependent Raman characterization, which is identified as the primary open validation gate. The companion members of the sub-genus — SrB4O7, MgAlBO4, CaB4O7, and Ca2SiB2O7 — extend coverage across a range of lattice symmetries and cation chemistries while retaining the fundamental BO4 network that provides the wide-gap electronic structure. SrB4O7 is a known nonlinear optical material with a large measured bandgap, providing experimental grounding for the borate-network electronic structure prediction. MgAlBO4 introduces a mixed-cation site that can tune the thermal expansion coefficient closer to silicon or silicon carbide substrates, relevant for heterogeneous integration. The calcium borates CaB4O7 and Ca2SiB2O7 offer higher-temperature processing stability and candidate compatibility with ceramic interposers. All members share the strategic property of being composed entirely of first-row and second-row main-group elements with no critical-mineral content, which is the defining selection criterion for this sub-genus. Dielectric-tensor DFPT simulations established the key permittivity figures for BPO4 and will need to be extended to the full set of sub-genus members as the program advances. The DFPT workflow used here captures both electronic and ionic polarization contributions, which is necessary for accurate prediction of interlayer dielectric performance at the low-frequency (DC leakage) and RF regimes relevant to 2.5D/3D packaging. No migration-barrier or NEB calculations are reported for this material family, which is appropriate given that ionic conduction is not the primary failure mode of concern for a borophosphate barrier dielectric; the relevant reliability metric is time-dependent dielectric breakdown (TDDB), which scales with bandgap and is indirectly supported by the 7.3 eV gap figure.

The primary commercial target is the advanced semiconductor packaging market, specifically the interlayer and barrier dielectric materials consumed by outsourced semiconductor assembly and test providers (OSATs) and by integrated device manufacturers running their own advanced packaging lines. This market segment — covering dielectric films used in 2.5D interposers, fan-out wafer-level packaging, and chiplet-integration stacks — is growing rapidly as the logic roadmap shifts from monolithic die scaling to heterogeneous integration. The total addressable market for dielectric materials in advanced packaging is estimated in the range of $0.5 billion to $1 billion annually, with growth driven by AI accelerator packaging, high-bandwidth memory stacks, and 5G/6G RF modules. These are estimates reflecting current market data; the figure should be treated as directional, not as a precise forecast. Beyond packaging, the radiation-hard electronics segment represents a distinct and higher-margin opportunity. Space electronics, nuclear instrumentation, particle physics detector infrastructure, and defense avionics all require dielectric materials that maintain insulating integrity under sustained particle bombardment. Wide-bandgap dielectrics are specified for these applications precisely because a large gap reduces the probability of ionization-track-mediated conduction. The addressable base here is smaller than the packaging market but commands premium pricing and long-term supply agreements. A licensee targeting both segments with a single material platform — which the borophosphate family enables — gains cost and qualification efficiency that competing rare-earth or transition-metal-oxide dielectrics cannot offer at equivalent supply security. The royalty logic for this type of materials patent is most naturally structured as a per-wafer or per-area license paid by the OSAT or IDM integrating the film, analogous to how specialty low-k dielectric precursor licenses have historically been structured. A secondary licensing path exists through the precursor chemistry and deposition-process supply chain, where chemical vapor deposition or atomic layer deposition precursor suppliers could take a sublicense. Given the earth-abundance driver, there is also potential interest from government-funded supply-chain resilience programs (CHIPS incentives, EU IPCEI) that might co-fund qualification work in exchange for domestic supply commitments, which could reduce the commercial licensee's cost of risk.

earth-abundant low-supply-risk wide-gap dielectric

supply-resilient alternative to RE/Nb/Ta-bearing members

The natural licensing counterparties are large OSATs — Amkor, ASE, JCET, and UTAC — that are under direct customer pressure from hyperscalers and IDMs to demonstrate supply-chain resilience in their materials procurement. These companies regularly take material licenses as part of process qualification agreements and have the engineering infrastructure to run thin-film coupon deposition experiments needed to close the open validation gate. A second tier of buyers includes advanced packaging equipment and materials suppliers — applied materials, Lam Research, and specialty ALD precursor chemists such as Entegris or Versum — who would seek to bundle the intellectual property into a process kit or precursor offering sold to OSATs. Defense primes and space system integrators (Northrop Grumman, BAE Systems, Lockheed Martin, and their tier-one electronics suppliers) represent a third buyer cohort specifically for the radiation-hard application, where long-term supply contracts for qualified dielectric materials are standard procurement practice and where the wide-bandgap property rather than cost is the primary selection criterion. Any of these buyers could structure the engagement as a development license (co-funding the thin-film coupon work in exchange for exclusivity in a defined application space) or as a standard royalty-bearing production license once the material is qualified.

The primary technical risk is the unresolved phonon picture at the zone boundary for BPO4, which means the dynamic stability verdict remains provisional until thin-film experimental data is in hand. This is a real uncertainty rather than a formality, and a licensing conversation must be honest about it. The mitigation path is clear and not expensive: ALD or sputtering deposition of a BPO4 coupon followed by Raman spectroscopy is a standard materials characterization experiment that a well-equipped OSAT or academic partner can run within a few months. If the zone-boundary soft mode proves to be a genuine surface or grain-boundary instability in thin-film form, the sub-genus still retains value through SrB4O7 and the calcium borate members, which have the same wide-gap borate topology and have not shown the same marginal phonon behavior. The second risk is commercial timing: there is no identified forced-substitution deadline (no regulatory date mandating the elimination of specific rare-earth dielectrics from packaging lines), so adoption will be pull-driven by supply-chain events — a critical-mineral shortage, an export control action on niobium or tantalum, or a customer mandate — rather than by regulatory push. This means the commercialization timeline is uncertain but the eventual demand, driven by structural supply-chain pressure, is directionally robust. A licensee taking a defensive position now, before supply disruption occurs, captures more favorable pricing than one reacting to a crisis.